Outside of city- and touring bikes as well as the occasional niche mountain bike, chains feature on the vast majority of transmission setups. Particularly when it comes to performance-oriented bikes, belts are a rare sight. This is also reflected in the products of mainstream drivetrain brands like Shimano or SRAM, all of which focus exclusively on chain transmissions in their performance ranges.
Do efficiency tests back up this narrow focus or is there a place for the belt drive on a performance bike? Let's have a look.
If you're just here for the results, click here.
In 2012, Friction Labs compared the friction of a Shimano CN7901 Dura Ace 10sp chain, a SRAM PC1091R Red 10sp chain and two Gates CDX belts. These tests were carried out in a single-speed rig (without a derailleur) equipped with chain-/beltrings and sprockets of near equal radius - 53/19 for the chain transmissions and 60/22 for the belt transmissions. All four setups were subjected to one hour run-in and the chains were cleaned and re-lubed with basic light bearing oil. The test rig subjected the transmissions to a steady 250 watts at 95 RPM.
In a later article, we will combine the findings of this comparison with the results of various tests involving derailleurs, gear hubs, different types of lubrication as well as the effects of contamination.
But, first things first: Let's start with singlespeed setups in lab conditions.
Belt transmissions require substantial preload tension for the belt to reliably stay on the beltring under use. When a rider pedals a belt drive, the tension created by their power input is added to the preload tension in the top span of the belt - the part of the belt that is being "pulled on" by the forward rotation of the beltring. However, that same amount of power-dependant tension is subtracted from the lower span of the belt. This means that the overall tension in belt transmission is entirely dictated by the preload tension, irrespective of power input.
At a theoretical preload tension of 0 lbs, a CDX belt would produce about 1.73 watts in frictional losses - and immediately skip off the transmission under pedaling.
For their tests, Friction Labs used a preload tension of 85 lbs, which was the recommended maximum back in 2012. This preload tension resulted in frictional losses of 3.93 watts. However, as of 2021, after a decade of advancements in frame stiffness and belt retention, Gates recommend a preload tension of around 43 lbs for demanding MTB setups. This is interesting since it would result in negative tension in the lower belt span at rider outputs beyond 200 watts, let alone sprinting efforts. But, given that these recommendations come directly from the manufacturer, we are going to assume that the introduction of belts with centerline tracks as well as the use of belt snubbers do indeed suffice to prevent the belt from skipping. So, from Friction Facts' measurements, we can deduce that this new preload recommendation would result in frictional losses of about 2.8 watts.
It's an entirely different story with chains, where the preload tension just amounts to a total of 3 lbs across both spans and frictional losses are dictated by power input. More acurately, frictional losses increase in a linear manner with power input.
Theoretically, a completely untensioned chain creates almost 0 watts in frictional losses. This changes once you introduce a derailleur to the setup - but since we're just comparing chains and belts at the moment, rather than complete drivetrains, we'll ignore this for now.
In a singlespeed setup, Friction Facts have measured the frictional losses of their 10spd chains to be 2.92 watts.
Up to this point, we've compared chains and belts as singlespeed setups in lab conditions. In this isolated setting, we can conclude that theoretically, a rider on a belt driven bike who sticks to Gates' current recommendation of maximum tension will have a performance advantage once their power output exceeds 240 watts.
At a more leasurely output of 150 watts, chain setups will run about 0.9 watts more efficient, whereas there is about a 0.4 watts difference at 200 watts.
But really, this is only half the story. To get a more conclusive picture, we need to take other factors into consideration:
Derailleurs and drivetrains warrant their own article, so we're sticking with singlespeed setups in this part of our series on chains and belts. However, we can tick off the first two items in that list by drawing on what we already know from our article on chain lubricants and contamination.
This is a topic which has largely been covered in a separate article, so instead of explaining all things lube and dirt here, we'll just jump to the conclusions - you can always head over to the article for all the details.
In the tests referenced above, all chains were lubed with light bearing oil. However, we know that we can save about 1.5 watts in dry conditions at a pedaling power of 250 watts when using state of the art wax based lubricants. This would make a chain 1.4 watts more efficient than a belt at that power.
While this pushes the rider power required for a belt to run more efficiently than a chain to unsustainable figures, we're still somewhat in the region of marginal gains. This changes once we introduce contamination - and this is were things become tricky. Even in slightly damp and dirty conditions, the efficiency of a chain drive train decreases by 1.66-4.33 watts. And this number climbs beyond 7 watts in rain all the way to up to 30 watts once we're talking full mudfest conditions.
The tricky bit is that we don't have any data on the efficiency of belts under similar conditions. But consider the following:
Belts don't require lubricants, hence dirt won't stick to them as badly as it does with chains, particularly those treated with wet lubricants.
Furthermore, chains consist of many articulating parts. Once the lubricant between the links (and rollers) gets contaminated, these links will have a harder time rotating around the pins keeping them together. This is a major source of contamination-induced friction that simply isn't present on a belt. Belts are influenced by contamination only because the beltring and sprocket teeth pushing into the belt needs to push out dirt to properly interface with the track.
So, while we don't have any numbers, it can be assumed that belts perform considerably better under harsh conditions - to the extent where the comparatively marginal advantage of freshly lubed chains under dry conditions will be quickly negated even in slightly wet weather or on prolonged offroad rides. Take this with a pinch of salt (as we can really only make educated guesses), but once you start riding through properly bad conditions or go beyond the required service interval of your chain lube, it wouldn't be surprising to see belts leaping ahead by over 5 watts at a pedaling power of 250 watts.
At this point, it's fair to say that if you're building a performance oriented singlespeed setup, you're almost certainly going to be better off with a chain drivetrain as long as you stick to the road and ride mostly in dry weather. A well maintained wax-lubed chain will run 1.4 watts quicker than a belt at a rider power of 250 watts in these conditions. This advantage grows at lower power ouputs by about 0.5 watts per 50 watts.
After a short ride in the wet or a short distance offroad race, both singlespeed transmission types are probably on equal footing due to contamination/degratation of chain lubrication.
On the other hand, if you're building a singlespeed rig for long distance offroad rides or can't service your drivetrain even after prolonged encounters with the rain, (all other things being equal), belts will likely have an efficiency advantage.
But what if one gear won't do it? In our next part of the series, we'll bring derailleurs and gear hubs into the mix.